Site connector (CLI/Newt 1.26) fails registration with “Site last hole punch is too old” while older Newt works on same network

未关闭
#178 3 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
4/5
预计耗时
3-5 天
新手友好度
38/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
go, linux
领域
networking

调研方向

Start by tracing the CLI site's registration and hole-punch flow, then compare it with the older Newt path; the payload does not name source files or tests. Check how the site identifies its network address and handles relay fallback, using the reported UDP ports and timeout logs as clues. Done means the CLI site registers and receives a working peer configuration, through hole-punch or relay.

由索引模型根据 Issue 内容生成。

描述

Describe the Bug

I am migrating from the Newt site connector to the now recommended Pangolin CLI site connector. The old Newt site works reliably. A newly created site using the current CLI binary on a Proxmox LXC container consistently fails to register.

Environment
  • OS Type & Version: Ubuntu 26.04 LTS (Pangolin server VPS) Debian 13 (Pangolin CLI LXC)
  • Pangolin Version: 1.26.0
  • Gerbil Version: 1.8.0
  • Traefik Version: 3.7.17
  • Newt Version: 1.20.0
  • Client Version: 0.21.0
To Reproduce

Creating a new CLI binary install on a Proxmox LXC container and attempting to connect to my self-hosted Pangolin server fails. Proxmox is on the latest 9.2 version, with tun access given to the LXC.

Inspecting server-side logs, I see:

Site last hole punch is too old; skipping this register.
The site is failing to hole punch and identify its network address with the server.
Can the site reach the server on its Relay UDP port?

From the client side, I see the following logs:

Websocket connected
Connecting to endpoint: pangolin.labnest.net
SendMessageInterval timed out after 16 attempts for message type: newt/wg/get-config
Initial reliable ping failed... failed to read ICMP packet: i/o timeout
(repeated ping timeouts)

UDP packets from the LXC reach my VPS on both ports 51820 and 21820 (confirmed with tcpdump and nc -u). I verified no firewall is blocking the ports.
When started with -native flag, the site does create a WireGuard interface and receives a tunnel IP, but never completes registration and never receives a working peer config.

I have tried fully deleting and recreating the site and get the same error. Setting a static port also doesn't help.
I compared ip addr and wg show between the working Newt and failing CLI sites. Both create an interface and get a tunnel IP.

AI Disclosure

Utilized Grok to troubleshoot error logs and attempt to get the connection working. Also used it to help me write this issue.

Expected Behavior

The CLI site should complete hole-punch (or fall back to relay) and receive a WireGuard configuration.

主要语言
Go
星标
70
派生
26
平均合并
4 小时 19 分钟
30 天内合并 PR
11

环境准备

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

fosrl/cli 的其他 Issue

查看 fosrl/cli 的全部 Issue

相似的 Issue

更多 Go Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。